Mechanisms and functions of ATP-dependent chromatin-remodeling enzymes

Abstract

Chromatin provides both a means to accommodate a large amount of genetic material in a small space and a means to package the same genetic material in different chromatin states. Transitions between chromatin states are enabled by chromatin-remodeling ATPases, which catalyze a diverse range of structural transformations. Biochemical evidence over the last two decades suggests that chromatin-remodeling activities may have emerged by adaptation of ancient DNA translocases to respond to specific features of chromatin. Here, we discuss such evidence and also relate mechanistic insights to our understanding of how chromatin-remodeling enzymes enable different in vivo processes.

Figure 1

Model of Isw2 Nucleosome Interactions The Isw2 complex represents a paradigm for nucleosome remodeler interactions, as the sites at which peptides derived from translocase lobe 1 and the HAND/SANT and SLIDE domains interact with nucleosomal DNA (black) have been determined by directed crosslinking (Dang and Bartholomew, 2007). The HAND/SANT SLIDE domain of Isw2 (gray) is modeled using the equivalent region of the Isw1 (Yamada et al., 2011). Two peptides from this region of the ISW2 complex (green and purple) crosslink to the bases shown in the space-fill of the same color. The RecA lobes (light-blue) are modeled using the structure of zebrafish Rad54 (Thomä et al., 2005) and the peptide shown in red crosslinked to the bases shown in red space-fill. The precise details of how each domain is docked are not known, as the specific amino acids that form crosslinks within each peptide are not known. Black dots indicate the positions of single-nucleotide gaps that interfere with nucleosome sliding (Zofall et al., 2006). Core histones are shown as yellow.

Figure 2

Mechanisms for Nucleosome Repositioning (1) Single-molecule measurements indicate that DNA is first removed from nucleosomes on the exit side (red) (Deindl et al., 2013). As a result, the intermediates in repositioning contain a deficit of DNA that has been measured as between 4 and 7 bp. This could be accommodated as a change in the conformation of the octamer (2), a reduction in DNA twist (3), or a combination of these. DNA is subsequently drawn into the nucleosome on the destination side (blue) (4), allowing the nucleosome to return to a more normal conformation 3 bp further along the DNA (5). This contrasts with previous models in which DNA was proposed to be drawn into the nucleosome prior to being removed (6). Note that, although DNA appears to enter and leave the nucleosome in 3 bp steps, these are likely to arise from three successive 1 bp movements of the remodeling enzyme.

Figure 3

Structural Models for Chd1 and Isw2 The chromodomains, translocase lobes (Hauk et al., 2010), and SANT/SLIDE DNA-binding domain (Sharma et al., 2011) of Chd1 are colored yellow, blue, and purple, as indicated in the schematic. The structure of linker sequences was crudely modeled based upon secondary structure prediction to indicate their scale rather than conformation. To the right, the helicase lobes are shown as space-fill, with the conserved DNA-binding motifs I, II, and III of lobe I and motifs V and VII of lobe II indicated in red. These conserved motifs are observed to be held in an open conformation that is likely to be inefficient for ATP-dependent DNA translocation. A similar model is shown for Isw2. In this, the HAND-SANT-SLIDE domain is modeled on Isw1 (Yamada et al., 2011), and the ATPase lobes are modeled using the structure of Zebrafish Rad54 in a configuration close to the closed conformation likely to be active for DNA translocation (Thomä et al., 2005). For both Chd1 and Isw2, accessory sequences contribute to the regulation of catalytic activity, and this may well involve changes in the alignment of the ATPase lobes. For example, the chromodomains of Chd1 and R93 of the ISWI protein (in red space-fill) confer negative autoregulation (Clapier and Cairns, 2012; Hauk and Bowman, 2011). This region also undergoes a conformational change upon DNA binding (Mueller-Planitz et al., 2013). In contrast the SANT-SLIDE domains of both proteins confer positive regulation (Hota et al., 2013; McKnight et al., 2011). DNA fragments bound to the SANT/SLIDE domains and modeled into the translocase domain are shown in red space-fill. However, it should be noted that, as the conformation of linker sequences (green) is not known, it is not possible to infer the orientation of the two bound DNA fragments.

Figure 4

Organization of Chromatin-Remodeling Enzymes with Respect to Transcribed Genes A schematic representation organization of chromatin-related factors with reference to the transcriptional start site (TSS). The genome-wide distributions of nucleosomes and posttranslational modifications to histones and RNA polymerase subunits reveal that many of these factors are organized with respect to transcribed genes (reviewed by Rando and Chang, 2009). More recently, it has become apparent that ATP-dependent remodeling enzymes also show distinct distributions with respect to transcribed genes where they influence nucleosome organization (Gkikopoulos et al., 2011; Yen et al., 2012). Furthermore, in some cases, the action of remodeling enzymes can influence the distribution of histone modifications and variants, whereas in other cases, modifications instruct the action of remodelers. This interplay between histone modifications and ATP-dependent chromatin modifications acts to sculpt the chromatin landscape on a genome scale and is likely to involve further integration with transcriptional elongation factors and additional factors acting to regulate chromatin organization, such as histone chaperones.